Learning Goals:
- Present arguments for why naturalization might foster or hinder
immigrants’ social integration.
- Explain the basic idea behind a regression discontinuity design
- Practice testing for implications of arguments when direct tests are
not possible
- Practice thinking through the generalizability of research findings
to “out of sample” populations
Arguments
Quickly write down:
- Through which mechanisms might naturalization foster social
integration?
- And why might granting naturalization too easily hinder social
integration?
Double Selection
- What’s wrong with simply comparing naturalized vs. unnaturalized
immigrants?

- How do Hainmüller et al. get around the first selection
problem?
- And the second?
Voting on your fellow citizens
Leaflet

But of course, people who win their referendums are still different from
people who lose them:
Who Gets a Swiss Passport?
—
So how does the research design help us to ensure that we are
comparing apples to apples?
Selection on Observables Strategy
In the first strategy we utilize the fact that we can measure and
control for all the applicant characteristics that were reported to
voters in the voting leaflets when they decided on the applicants and
therefore rule out omitted variable bias. In contrast to the situation
where an immigration official decides on the applicants based on
information that is unobserved to the researcher, here we do observe all
the relevant applicant characteristics that were reported to voters who
decided on each request. Once we adjust for the reported characteristics
and compare applicants who applied in the same municipality, in the same
time period, have the same gender, country of origin, marital status,
number of children, education, occupational skill, years of residency,
assessed integration level, and language proficiency, such matched
applicants are observably equivalent to voters, and therefore voters
cannot systematically discriminate between applicants based on their
unobserved characteristics. Therefore, among such observably equivalent
applicants who are matched on the characteristics that voters see on the
leaflets, who wins and who loses is not driven by systematic differences
in the integration potential of the individual immigrants, but by
idiosyncratic shocks that affect the aggregate vote outcomes such as
what else appeared on the ballot or the weather on the day of the
referendum.
- What does that mean?
- How is this different than simply using controls in an observational
study?
RD Design
Again, the goal here is to create similar treatment and control
groups.
But rather than relying on statistical controls, we rely on the
assumption that people who barely lost their referendum
are, on average, the same as people who barely won.
Of course, still need to check whether this assumption holds:
Covariate Balance
—
RD Result
—
You try it:
It is very commonly found in surveys that people with a university
education are more tolerant of diversity and have more pro-immigrant
attitudes.
One hypothesis for this relationship is that attending university
causes people to become more tolerant.
- Explain why a simple comparison of people with vs. without
university education might be biased.
- How might you better test this hypothesis?
- Hint: consider that in some countries, the only way to attend
university is to sit for – and pass – a national entrance
examination.
Main Results

How should we interpret this result?
- Winners increase integration efforts: citizenship
provides naturalized immigrants with the identity, incentives,
recognition, and resources to increase their long-term social
integration
- Losers decrease integration efforts: applicants who
are denied became more alienated from Swiss society than they would have
become had they never applied for naturalization in the first
place.
Testing Implications
Note that we cannot directly test the alienation mechanism: that is,
we should not compare people who lost their referendum against people
who never applied for citizenship in the first place (even controlling
for observable differences).
(BTW, can you explain why not?)
Instead, we can make progress by reasoning by implication:
considering outcomes that would only be predicted by one mechanism, but
not the other.
An Example You Already Know: Statistical vs. Taste-based Discrimination
Imagine we conduct a correspondence study on housing
discrimination.
Suppose that:
- landlords are only concerned about tenants’ ability
to pay the rent.
- minority renters really are, on average, poorer than majority
renters.
We run an experiment where we send a short email to landlords with
the signature:
- Michael Fischer
- Mohammed el-Fatih
- Prof. Dr. Michael Fischer
- Prof. Dr. Mohammed el-Fatih
An implication of statistical (but not taste-based) discrimination is
that discrimination would disappear (or diminish) when signing with
“Prof. Dr.”
An implication of taste-based (but not statistical) discrimination is
that signing with “Prof. Dr.” should have no effect on discrimination
rates.
You try it

American university students commonly have the experience that the
football players in their classes say really dumb things.
Of course, the standard explanation for this is that
football players weren’t selected to attend university based on their
academic merit, but rather on their athletic skills (and ability to
generate $ for the school). In other words, football players say dumb
things in class because they are dumb.
But other explanations are also possible:
- Limited Time: There is limited time in a day, so
when a person engages in a very time-consuming activity (such as playing
semi-professional sports), it takes time away from other very
time-consuming activities such as studying. Given limited study time,
football players are less prepared academically, and therefore are more
apt to say dumb things.
- Jealosy: We are jealous of others’ success. When we
are jealous, we subconsciously lower our evaluation of that person’s
performance in other areas. So we think
football players say dumb things in class (when really, they do
not).
Your task:
Can you come up with any observable implications of these theories
that would allow you to distinguish them from the standard explanation
(and from each other)?
In other words, can you think of outcomes that would only “fit” one
theory, but not the others?
Implications of Alienation
If applicants become alienated because their applications have been
denied, then we expect:
- they would develop a higher level of distrust of the local
authorities who handled the applications and did not avert the
potentially discriminatory rejections.
- they would develop a higher level of distrust of the judicial system
more broadly because the courts did not overturn a discriminatory
rejection upon appeal.
- them to grow more distrustful of other people in their community
given that a majority voted against their application.

“Out of sample” inferences
Hainmüller et al have answered a scientific question: naturalization
fosters immigrants’ social integration (within the Swiss context).
But suppose you are a politician. You have the following policy
question: can we improve social integration by making naturalization
easier (e.g. by lowering residency, language and employment
requirements)?
An ideal experiment: * another country similar to CH (same immigrant
groups, same level of xenophobia, etc.) * same referendum system in
place, during the same time period * but easier to get on the ballot
We then replicate the experiment in country X and compare the effects
of naturalization there vs. in CH.
Of course, country X doesn’t exist! We are asked to make an inference
about what would happen to an “out-of-sample” population.
Think about two groups:
- A: People who would apply if eligible, and are in
fact currently eligible
- B: People who would apply if eligible, but are not
eligible under the current regime
- How does the “treatment” likely work? (Look back to what you listed
at the beginning of class)
- Are these effects likely to be weaker or stronger for people in
Group B relative to people in Group
A?
Answering these questions should help you to reason “out-of-sample”
and offer some (principled) advice to the politician.